Page 1 of 10 2015-PPIC-0187

SIGNATURE ANALYSIS FOR ON-LINE MOTOR DIAGNOSTICS

Ian Culbert John Letal

Senior Member, IEEE Qualitrol-Iris Power
Qualitrol-Iris Power 3110 American Drive
3110 American Drive Mississauga, ON
Mississauga, ON Canada
Canada

Abstract - Stator current signature analysis on
induction motors is a proven method for diagnosing rotor
squirrel cage winding defects and air gap eccentricity
problems. With this technology, specific frequency
current components can be identified as an indication of
cage winding defects as well as a non-uniform gap
between the stator and rotor. Because this data is
generally collected periodically, it is important to identify
these components as early as possible. These trendable
parameters can then be monitored more often to avoid in-
service failure. With the application of new processing
technologies, the ability to identify these critical current
signature frequency components and trend the
deterioration they indicate has improved. This allows for
maintenance activities to be scheduled earlier and
performed prior to failure avoiding costly motor
component damage and unplanned downtime.
term operation with broken rotor bars can also cause heat
damage (Fig. 2) to the rotor core because the current normally
carried by broken bars or shorting rings is routed through
adjacent bars and the core. Large air pockets (Fig. 3) in die
cast rotors and broken bars cause non-uniform bar expansion
which will result in rotor unbalance and high vibration. These
high vibration levels at rotational speed can cause premature
bearing wear. If the broken bars lift out of the slot or with
unbalanced magnetic pull, the air gap eccentricity will be
affected and may cause the rotor to rub with the stator (Fig. 4)
causing catastrophic winding failure. Air gap eccentricity can
produce high twice line frequency vibration at destructive
forces. To avoid such problems the frequencies of interest
must be identified as early as possible and trended so that
corrective action can be taken prior to failure.

Index Terms - stator current signature analysis, rotor

squirrel cage winding, broken rotor bars, air gap eccentricity,
induction motor, condition based maintenance.

I. INTRODUCTION
Condition based maintenance is a strategy to detect and
identify rotating machine problems prior to failure. By
collecting various data from a machine, operators or
maintenance personnel can determine if a machine is
exhibiting certain patterns that may be an indication of
degradation of certain motor components. With this
knowledge, repairs can be scheduled in a manner that will
maximize plant process and still avoid costly unplanned
Fig. 1 Failed brazed connection [3]
downtime for unscheduled outages and repair. Generally, if an
in-service failure does occur, the repair costs will be greater
due to the possibility of collateral damage than if the
maintenance is scheduled appropriately [1-2].
For induction motors there are a variety of tests that can be
used to assess the condition and drive maintenance efforts.
One such online test, meaning data is collected during normal
operation, which has been used for more than a decade is
stator current signature analysis. It is a proven method for
diagnosing rotor cage winding and air gap eccentricity
problems by analyzing the frequency spectrum of the stator
current signature. Cage winding faults including broken rotor
bars (Fig. 1) are not necessarily cause for immediate concern,
but there can be serious secondary effects on the machine
operation. With these types of rotor problems the motor can
continue to operate, but with reduced torque. The rotor slip Fig. 2 Rotor core burning from broken bars
and line currents will increase for the same load. Any sparking
that results is a safety hazard in explosive environments. Long

978-1-4799-7114-5/15/$31.00 © 2015 IEEE

 2015-PPIC-0187
Fig. 3 Die cast rotor with air voids (courtesy of EASA)
Page 2 of 10
II. FAILURE MECHANISMS IDENTIFIED WITH SIGNATURE
ANALYSIS
Frequency components of motor stator current can be
collected by sampling the load current on one phase. If a cage
winding or air gap eccentricity problem is present, this load
current will have a certain pattern. Specific fault frequency
components can be identified and trended as long as there is
an accurate calculation of the rotor slip frequency. Slip speed
(in rpm), calculated with (1), is dependent on machine design
and operational load. Per unit rotor slip “s” can be calculated
by measuring rotational speed with a stroboscope or with a
laser tachometer. Some instruments that collect current
signature data have advanced algorithms which calculate
operating slip based on load current. More load results in
higher slip, and therefore, lower rotor turning speed [3].

 



 



 (1)

 



Fig. 4 (a) Rotor surface damage due to rotor rubbing stator
Fig. 4 (b) Stator core damage due to rotor rubbing stator
With the advancement of signal processing capabilities,
frequencies related to these fault conditions can be identified
earlier than even 5 years ago. The speed of analog to digital
converters available today has vastly improved. This allows
for a much higher sampling rate so that an appropriate
frequency range can be used to identify the cage winding and
air gap eccentricity problems described above. Higher
dynamic range is available so that these frequencies can be
identified and trended at the onset of a problem. By gaining
insight to this condition-based knowledge even earlier than
before, maintenance planners now have the ability to assess
cage winding and air gap eccentricity problems earlier and
schedule repair work accordingly prior to failure.
978-1-4799-7114-5/15/$31.00 © 2015 IEEE
Where;
s per unit rotor slip
A. Cage Winding Failures
With an accurate detection or calculation of rotor slip, fault
frequencies can be determined to assess the condition of
induction motor rotor related conditions. Broken or cracked
rotor bars, shorting rings and the connections between them
in fabricated rotor windings as well as air pockets in die cast
rotor windings can be identified with one such fault frequency.
A 3-phase magnetic field with the same number of poles as
the stator field at the rotor slip frequencies is produced by the
rotor currents. If rotor asymmetry occurs due to cage winding
cracks or breaks, there will also be a resultant backward
rotating field at slip frequency with respect to the forward
rotating rotor. This produces a stator current with a frequency
(1-2s)f1 about the fundamental supply frequency. Because the
rotor is constantly catching up to the stator magnetic field, the
result on the rotor is an oscillation that generates a current
component with a frequency of (1+2s)f1 (2) about the
fundamental supply frequency.
  1  2 (2)
where
fsb twice slip sidebands
f1 supply frequency
The magnitude of the currents generated at the frequencies
of these sidebands increases due to [1, 3]:
• broken rotor bars
• higher motor loads
• developing additional cage winding defects.
The difference in amplitude between the fundamental
supply frequency and the twice slip sideband currents gives
rise to the need to identify small currents that are sometimes
200 times less than the fundamental, i.e. if the motor is
drawing 100A the oscillating current at the twice slip sideband
may only be 0.5A. This is the reason that a log scale (dB) is
generally used for current signature analysis and also drives
the requirement for a high dynamic range in signal
Page 3 of 10 2015-PPIC-0187
processing, especially since early identification of these
frequencies can help schedule maintenance activities.
Broken rotor bars and other cage winding faults can be
identified in a stator current signature with these classical
twice slip sidebands if they are 45dB (or less) below the
fundamental supply frequency [3]. The smaller this difference
is, the greater the cage winding damage and in fact, if the
number of rotor bars are known, it is possible to calculate how
many are broken, as in Fig. 5 where 3 or 4 broken bars are
identified.




Fig. 5 Current signature cage winding spectrum
B. Air Gap Eccentricity
The air gap is the separation between the rotor and stator
in a motor which conducts the magnetic flux from the stator
and links it to the rotor inducing a radial force, or torque, on
the rotor causing it to turn. If this gap is uniform there will be
no change with rotor rotation. If the center of the rotor is
displaced from the center of the stator bore, the minimum air
gap will stay in the same position and the result will be static
air gap eccentricity. Over time static air gap eccentricity will
cause the rotor shaft to bend resulting in dynamic air gap
eccentricity. When the rotor shaft is bent, the position of the
minimum air gap will change. It is important to note that if a
motor has air gap issues there is usually a combination of
static and dynamic eccentricities. Not only does the rotor have
to be machined straight to remove the dynamic eccentricity,
but the bearing positions need to be adjusted to remove the
static eccentricity.
The fault frequencies of interest when analyzing motor air
gap are calculated with (3) which is a combination of static
and dynamic forces. In particular, the stator rotor slot passing
components are spaced at twice the fundamental supply
frequency and the peak with the highest magnitude should be
trended. Dynamic eccentricity can be identified with rotational
speed sidebands around the highest magnitude rotor slot
passing component [4].


Rotor slot passing components indicate static eccentricity
when the dynamic eccentricity index is 0. If there is dynamic
eccentricity as well, the dynamic eccentricity index will be ±1
generating sidebands around each air gap eccentricity
frequency and the rotation speed frequency (4). The spectrum
when the dynamic eccentricity index is 0 will show the rotor
slot passing components spaced at twice fundamental supply
frequency, first part of (4). If dynamic eccentricity exists, which
is nearly always the case, the slot passing frequencies will
have sidebands at rotational speed frequency, second part of
(4). Some monitors can even determine the rotor slip from
motor nameplate information and measured load current. The
static and dynamic air gap eccentricity components can also
be determined and identified with pattern recognition
algorithms for analysis eliminating the need for experts who
have a firm grasp of the mathematics introduced above.
Static Dynamic
 =  [ R (   #
] ±  [ ] (4)
The criterion for air gap eccentricity severity is based on the
average difference between the highest magnitude rotor slot
passing frequency component and the rotational speed
frequency components. In general if this average difference is
15 to 25 dB then a problem related to air gap eccentricity
exists and must be investigated. This is illustrated by the
following example of an air gap eccentricity analysis
performed on a 2700 hp (2014 kW), 4.0 kV, 60 Hz, 6-pole
Flue Gas Compressor Motor in a combined cycle power plant.
The air gap eccentricity current versus frequency spectrum
around the highest rotor slot passing frequency from this
motor is shown in Fig. 6. From this, it can be seen that the
highest magnitude rotor slot passing frequency is -69 dB and
the average rotational speed sidebands is -83.25 dB (average
of -79 and -87.5). The resulting difference is 14.25 dB which is
unacceptable.
Fig. 6 Current signature air gap eccentricity spectrum
         !
" (3)

Since such excessive air gap eccentricity will produce much

higher than normal 2 x power supply frequency (120 Hz)
where bearing housing radial vibration levels the most recent motor
fec air gap eccentricity current frequencies vibration levels were analyzed. As can be seen from Fig. 7
R number of rotor bars this analysis did in fact show a 120 Hz vibration magnitude of
nd dynamic eccentricity index (0 or ±1) 4.537 mm/s rms which was assessed to confirm the current
p number of pole pairs signature analysis indicating poor air gap eccentricity.
nωs odd integers (1, 3, 5 . . .)

978-1-4799-7114-5/15/$31.00 © 2015 IEEE

 2015-PPIC-0187
Fig. 7 Motor Bearing Radial Vibration Spectrum
C. Trending
Once cage winding and air gap eccentricity fault
frequencies have been identified, they should be trended at
approximately the same load so that an increase can correctly
be attributed to degradation in the condition of the motor. In
the case of cage winding faults, a broken bar factor is a good
indicator to assess the machine and schedule an outage
appropriately. If the number of bars is not known, which is
often the case, a severity indicator should be based on the dB
difference between the fault frequencies and fundamental
supply frequency. Similarly, the average difference between
the highest magnitude rotor slot passing frequency
component and the rotational speed frequency components
should be trended for air gap eccentricity issues. Generally
measurements should be taken at least once every six
months based on the criticality of the motor to the process. If
these trendable parameters start to show warning or problem
levels, then the frequency of measurements should be
increased to even at least once a month until repair work can
be carried out, as is shown in Fig. 8.
After repair
Fig. 8 Broken rotor bar factor trend
III. TECHNOLOGY ADVANCEMENTS FOR EARLY DETECTION
In general, analog to digital converters (ADCs) are selected
based on application requirements. For industrial
measurement, sigma-delta ADCs satisfy the requirements of
high resolution with a reasonable sampling rate to effectively
collect stator current signature data. As well, small output
978-1-4799-7114-5/15/$31.00 © 2015 IEEE
Page 4 of 10
voltages can be digitized directly [4]. With suitable sampling
rates and digital filter bandwidths, sigma-delta ADCs allow for
supply frequency rejection, meaning no additional signal is
introduced due to processing. This is necessary for assessing
cage winding faults as severity is based on the difference
between a relatively large supply frequency and small twice
slip sidebands.
In the 1950s the basic concepts of sigma-delta ADCs were
developed and by the end of the 1960s they were well
understood [4]. Unfortunately, the digital filters required to
effectively use these converters were not available until the
late 1980s limiting the practicality. Currently, high dynamic
range (up to 24-bit) and high sampling rate (up to 144 kHz
bandwidth) sigma-delta ADCs are available on the market
exceeding the requirements and providing much more
effective current signature analysis.
High dynamic range allows for the detection of current
signature fault frequencies riding on a relatively high
amplitude supply frequency. Typically a greater than 80 dB
dynamic range is required to process air gap eccentricity.
Currently, sigma-delta ADCs with 24-bit resolution allows
greater than 110 dB signal to noise ratio resulting in earlier
detection of these low amplitude frequencies so that
preparation can be made well ahead of machine failure. On a
logarithmic scale, this 30 dB difference is quite substantial and
is especially important for early detection of problems.
In order to capture both cage winding and air gap
eccentricity fault frequencies with a single recording, a high
sampling rate is required. A reasonable full scale range of 7
kHz will provide a usable range of 3 kHz when considering
Nyquist sampling theorem so that the frequencies of interest
are recorded. Obviously, one recording during data collection
intervals is preferred from a time-management point of view.
When considering cage winding faults, this high resolution is
also an important characteristic for collecting data on motors
with low slip or on low speed motors in order to separate the
supply frequency from the fault frequencies.
IV. CONCLUSION
Stator current signature analysis is a proven method to
detect cage winding faults and air gap eccentricity issues on
induction motors. It is a test that can be performed while the
machine is in operation and can be used to confirm the cause
of high vibration that may be related to rotor problems or the
interaction between the rotor and stator. The fault frequencies
that are identified with current signature analysis are trend-
able and an increasing trend is an indication of a degrading
condition.
Cage winding failures are prominent in motors that directly
drive high inertia loads such as induced draft and forced draft
fans, motors that have frequent starts and stops, and motors
with poorly manufactured cage windings. For these machines
it is critical that cage winding fault frequencies, i.e. the twice
slip sidebands are identified with current signature analysis so
that they can be identified prior to failure.
Air gap eccentricity results from poor assembly such that
the rotor is not concentric with stator core bore, the rotor shaft
is bent, and/or the sleeve bearings are worn. The potential for
these issues necessitates a system that can accurately
calculate complicated fault frequencies and identify them early
Page 5 of 10 2015-PPIC-0187

enough to schedule an outage to avoid costly collateral

damage related to air gap eccentricity.
Recent technological advancements in analog to digital
converters has allowed for the detection of fault frequencies
that can identify cage winding faults and air gap eccentricity
issues earlier. With early detection of these faults, plant
personnel have more time to schedule and plan outages to
optimize their process.

V. REFERENCES
[1] P. Tavner, L. Ran, J. Penman and H. Sedding,
“Condition Monitoring of Rotating Electrical Machines,”
London UK: The Institution of Engineering and
Technology, 2008.
[2] G.C. Stone, I.A. Culbert, E.A. Boulter, and H. Dhirani,
Electrical Insulation for Rotating Machines - Design,
Evaluation, Aging, Testing and Repair – Second Edition,
Hoboken NJ: John Wiley and Sons Inc., 2014.
[3] I. Culbert and W. Rhodes, "Using Current Signature
Technology to Reliably Detect Cage Winding Defects in
Squirrel Cage Induction Motors,” IEEE PCIC Paper No
PCIC-2005-11.
[4] W T Thomson, D Rankin and D G Dorrell: “On-line
Current Monitoring to Diagnose Air gap Eccentricity – An
Industrial Case History of Large HV, 3-Phase Induction
Motors”, IEEE Transactions on Energy Conversion, Dec.
1999.
[5] W. Kester, “Which ADC Architecture Is Right for Your
Application?,” Analog Dialogue 39-06, June 2005.

VI. AUTHORS’ INFORMATION

Ian Culbert has been a Rotating Machines Specialist at

Qualitrol-Iris Power since April 2002. Before joining Iris Power
he was a motor and small generator specialist with Ontario
Hydro/Ontario Power Generation from 1977 to 2002 and prior
to then a motor designer with Parsons Peebles, Scotland and
Reliance Electric, Canada. Ian is a Registered Professional
Engineer in the Province of Ontario, Canada and a Senior
Member of IEEE. He has co-authored two books on electrical
machine insulation design, evaluation, aging, testing and
repair and been principal author of a number of Electric Power
Research Institute reports on motor repair. Ian has also co-
authored a number of papers on motor electrical component
on-line and off-line motor diagnostics testing.

John Letal is a Rotating Machines Engineer at Qualitrol-Iris

Power responsible for supporting rotating machine
mechanical monitoring. Prior to Iris, he spent most of his
career as a Field Service Engineer troubleshooting large
rotating equipment using such tools as current signature and
vibration analysis. He was also involved in the implementation
and execution of vibration analysis programs. John holds a
Bachelor of Science degree in Manufacturing Engineering
from the University of Calgary and is registered as a
Professional Engineer.

978-1-4799-7114-5/15/$31.00 © 2015 IEEE

 2015-PPIC-0187 Page 6 of 10

Fig. 1 Failed brazed connection [3]

978-1-4799-7114-5/15/$31.00 © 2015 IEEE

Page 7 of 10 2015-PPIC-0187

Fig. 2 Rotor core burning from broken bars

978-1-4799-7114-5/15/$31.00 © 2015 IEEE

 2015-PPIC-0187 Page 8 of 10

Fig. 3 Die cast rotor with air voids (courtesy of EASA)

978-1-4799-7114-5/15/$31.00 © 2015 IEEE

Page 9 of 10 2015-PPIC-0187

Fig. 4 (a) Rotor surface damage due to rotor rubbing stator

978-1-4799-7114-5/15/$31.00 © 2015 IEEE

 2015-PPIC-0187 Page 10 of 10

Fig. 4 (b) Stator core damage due to rotor rubbing stator

978-1-4799-7114-5/15/$31.00 © 2015 IEEE